GPU usage

[Angelo Rodriguez]

hi there,
is it possible to detect if the DNC is using the GPU when is running the train.py
thanks

[dm-jrae]

Well you can log the device placement of ops in your graph:

https://www.tensorflow.org/tutorials/using_gpu#logging_device_placement

namely using

tf.Session(config=tf.ConfigProto(log_device_placement=True))

this prints out all op -> device placements. If the device ends in /gpu:0 (or /gpu:X where X is some integer) then it has been placed on the GPU.

[Angelo Rodriguez]

this is my Output:
Device mapping: no known devices.

any advice please?
thanks

[dm-jrae]

There is probably a problem with TensorFlow locating your nvidia driver, e.g. this is discussed in: drorlab/tf-singularity#3. Have you had any luck with this?

[Angelo Rodriguez]

at last i manage to find out a solution on this post:

https://github.com/deepmind/sonnet

all precedure fix it in this nutshell:

cd sonnet/tensorflow
git checkout r1.1
./configure
bazel build --config=opt --config=cuda //tensorflow/tools/pip_package:build_pip_package
bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
sudo -H pip uninstall tensorflow
sudo -H pip install /tmp/tensorflow_pkg/tensorflow-1.1.0rc1-cp27-cp27mu-linux_x86_64.whl
cd ../
bazel build --config=opt //:install
bazel-bin/install /tmp/sonnet_pkg
sudo -H pip uninstall sonnet
sudo -H pip install /tmp/sonnet_pkg/sonnet-1.0-py2-none-any.whl
python -c 'import sonnet as snt; import tensorflow as tf; print snt.resampler(tf.constant([0.]), tf.constant([0.]))'
cd ..
$ cd ~/
$ python

import sonnet as snt
import tensorflow as tf
snt.resampler(tf.constant([0.]), tf.constant([0.]))